Professional Documents
Culture Documents
REVIEW
The New Theory of Foreign Direct Investment: Merging Trade and Capital Flows
Katheryn N. Russ
University of California, Davis.
Foreign direct investment is a form of international capital ows [p. 8]. Assaf Razin and Efraim Sadka, Foreign Direct Investment: Analysis of Aggregate Flows. Princeton, NJ: Princeton University Press, 2007. . . . capital ows are important and interesting, but I also believe that they can be largely separated from the real decisions about the location of production and the direction of trade [p. xv]. James R. Markusen, Multinational Firms and the Theory of International Trade. Cambridge, MA: MIT Press, 2002.
I. Introduction
What is the origin and purpose of a multinational rm? How is it different from its native counterparts? In what countries and industries should one expect to see more of them? What is their impact on the pattern of trade and the distribution of wealth across countries? With the enormous growth of foreign direct investment (FDI) in the last 25 years, these questions have found voices in two divergent literatures, both rooted in the general
r 2009 Blackwell Publishing Ltd. 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA
108
Katheryn N. Russ
equilibrium modelling of the 1950s. Mundell (1957) rst mathematically modelled cross-border capital ows in a HeckscherOhlinSamuelson framework,1 simultaneously illustrating the two ideas underpinning these separate views of FDI in studies conrming or contradicting them for ve decades afterward: (1) that ows of capital and goods are substitutes and (2) that all else equal, capital ows should be related to relative endowments when there are barriers to trade, owing from capital-abundant countries to capital-poor ones. Hymer (1976) challenged this view by observing that cross-border ows of capital were likely to take place between countries with similar capital-tolabour ratios and within industries characterized by imperfect competition or fragmented production strategies.2 In the 1980s, it became the work of theorists like Ethier, Grossman, Helpman, Markusen and Razin to pinpoint exactly why FDI differs from the way Mundell and neoclassical growth models envisioned it, as it became clear that Hymer was right: FDI was increasing, but between rich countries and in tandem with intrarm trade. Some of these theorists, like Markusen, chose to use a trade-based, rmfocused approach, elucidating the structure of xed costs,3 relative country size and endowments, and preference conditions that compel rms to expand operations overseas. Others, like Razin, soon after with Sadka, chose a capital ows angle, embedding sunk costs,4 taxes and other distortions as fulcrums to analyse the behaviour of aggregate cross-border capital ows and their implications for country welfare.5 In Markusens text, Multinational Firms and the Theory of International Trade and Razin and Sadkas treatise, Foreign Direct Investment: Analysis of Aggregate Flows, these giants in the eld bring us up to date and carry us forward into the next generation of modelling, simulation and empirical analysis of the multinational enterprise (MNE). In particular, Markusen gives us a roadmap to match theories of the rm with observed aggregate ows, while Razin and Sadka derive new microeconomic fundamentals to formalize the more descriptive
1
Heckscher and Ohlin conceptualized a world, modelled mathematically by Samuelson, where free trade in goods generates welfare gains between countries with different capital-tolabour ratios, as well as the equalization of wages and capital rental rates across trading partners. Feenstra (2004, pp. 31, 459, 470) provides a complete description and listing of the articles that constitute the foundation of the HeckscherOhlinSamuelson framework. For a more detailed discussion of Hymers contribution to research on FDI, see Bhagwatis review of Hymers work reprinted in Feenstra (1983). Fixed costs are overhead costs that do not vary with the volume of production.
3 4
Sunk costs are costs incurred before a rm knows the precise level of net prots it will earn. They make investment decisions sensitive to uncertainty. See Russ (2007) for a detailed discussion of the literature on FDI and cross-border capital ows in the 1980s.
109
characterization of FDI common in international nance. To do so, the authors begin at very different places, but use tools with more overlap than one might expect. They leave the reader with whispers of a mechanism to reattach locational choice and intrarm trade ows with a rigorous theory of capital ows, as well as a plentitude of promising direct suggestions and methodologies for other fertile avenues of future research. Below, I discuss the way that their philosophies are manifest in these helpful manuals for modelling and empirics, as well as how their narratives relate to recent trends in the literature.
Chapter 5 is based on Markusen and Venables (1998) and Chapter 6 on Markusen and Venables (2000).
r 2009 Blackwell Publishing Ltd
110
Katheryn N. Russ
varieties (in oligopoly due to decreasing markups and in monopolistic competition due to an intrinsic love of variety itself ), so given the increasing returns to rm scale and the cost of shipping exports, some rms from each country will be able to prot from investing abroad. A larger market arising either by assumption or through eliminating restrictions on international trade and investment lowers the markup and increases rm scale in the oligopolistic model but has no markup or scale effect in the monopolistically competitive environment.7 He thereby emphasizes the role of industrial structure as an issue of central importance in measuring the determinants and welfare effects of FDI ows. Razin and Sadka also offer several different models that produce two-way ows of FDI between countries with similar endowments of capital and labour. However, instead of proximity, they propose a Ricardian engine of comparative advantage in investing. The approach is highly innovative in that it supposes comparative advantage in investing rather than in production, but is well supported by a rich empirical literature on creamskimming8 and the positive correlation of foreign takeovers with target rms post-takeover productivity. In one model, they assume that upon entry, each rm gets a noisy signal about how efcient its manufacturing process is, so that the owner is not certain what the optimal level of capital investment might be. Every rms true productivity lies within a xed interval from the signal it observes. The size of this interval is known and identical across rms. At this point, the original owner can hold onto the rm or sell it to a domestic or foreign investor. Any owner intending to produce in the following period must decide whether to pay a screening cost to nd out the true productivity of the rm and adjust its capital stock to the optimal level, or to simply retain the initial level of rm capital exogenously endowed upon entry. Naturally, only owners observing a signal above an endogenous threshold level will choose to bother screening and adjust the capital stock through additional investment. Foreign investors incur a lower cost when screening rms for a potential takeover in one of two or more industries. Therefore, each country is somewhat specialized in the sense that it has expertise that helps its residents to discern the true productivity of rms worldwide in a particular industry relatively cheaply, a type of Ricardian comparative advantage. A lower screening cost increases the value of a potential target for foreign investors,
7
The contrast in market structure is similar to that in Krugman (1979), which allows for variable markups, versus Krugman (1980), which imposes a constant markup in a monopolistically competitive environment. Cream-skimming describes the propensity of foreign acquirers to choose the most productive local rms as targets for takeovers.
111
allowing them to bid more than domestic investors to acquire a rm in the industry of foreign expertise. It seems quite plausible to imagine Germany better able to pick winners in the US pharmaceutical (or beer or auto) industry while the United States is able to cherry pick among targets in Germanys retail sector. Thus, one might see two-way ows as Germany invests in US pharmaceutical rms and the United States in German retail chains. Like Ricardian gains from trade in the traditional sense, there are real gains arising from comparative advantage in this model. The presence of foreigners with a relatively low screening cost for any industry has the real consequence of increasing nal output and improving economic efciency by having more rms acquired and screened for potential optimization of their capital stock than would be the case under capital controls. The previous model is introduced in partial equilibrium, with the two-way ows suggested as a reasonable extension by the authors. It stops short of explaining why capital does not ow from capital-rich to capital-poor countries as the Mundell and older neoclassical models would suggest. To do this, the authors explicitly construct a general equilibrium model,9 retaining the assumption that rms are heterogeneous but for simplicity removing any uncertainty about their idiosyncratic efciency levels. The screening cost is now just a xed setup cost, but due to superior expertise in management or R&D, it is again lower for foreign rms than domestic rms in one industry, again producing two-way ows. Capital in the model is perfectly mobile between countries but labour is not. The result is that the returns to capital can equalize across countries with different relative labour endowments, squelching further capital ows despite a wedge persisting between the wage rates. This is the explanation the authors offer for why low-wage countries remain capital-poor, amending traditional theories of capital ows. Clearly, the philosophical underpinnings of these two sets of models differ. In particular, Markusen is quite clear in stating at several points that FDI ows are not about transfers of physical capital, refraining from including any representation of capital in his production technologies, whereas Razin and Sadka use capital ows as a conceptual starting point. Markusen uses sophisticated models of competition to motivate two-way ows of FDI, Razin and Sadka appeal to a Ricardian logic. Markusen explains small ows to developing countries by pointing to their scarce supplies of skilled labour; Razin and Sadka suppose it is due to an excess of
9
It is a general equilibrium model in the sense that there are labour market-clearing conditions and endogenous wage rates. It is not clear how consumers behave or whether there is any output in the rst period that can be used to expand (endogenously) the capital stock in the second. I assume that there is no output in period one.
r 2009 Blackwell Publishing Ltd
112
Katheryn N. Russ
(undifferentiated) labour. Markusen rigorously models free entry and ex ante identical rms, while Razin and Sadka assume an exogenously xed number of heterogeneous rms. Markusen thinks about rms deciding whether to invest abroad. Razin and Sadka think about investors of an unspecied nature are they individuals, rms, institutions? It does not matter. Nonetheless, both theoretical approaches rely on the conjecture that it is some special technical or managerial know-how that generates the two-way ows of FDI between industrialized countries, which are the dominant feature of any objective description of the data. Markusen and other microlevel theorists envision the know-how as a non-rival resource yielding rmlevel economies of scale that encourage expansion through the establishment of overseas plants. Razin and Sadka and macro-nance theorists call it intangible capital. By any name, it can be used to improve the productivity of multiple plants simultaneously at relatively low cost (compared with the costs of starting a new rm or making investment decisions without this special resource) and can be generated through specialization, learning, R&D or innovation. In addition, the source of nancing for the investment be it mutual funds, bank loans, equity issues in overseas stock markets is indeterminate in both sets of models, set aside as an issue separate from the economically meaningful decisions of how to manage the rm. There is a fundamental meeting of the minds on these key issues. Notwithstanding, the authors of the two works are interested in a somewhat different set of stylized facts. First, Markusen considers the potential for intra-industry FDI ows to be important in matching any model to the data, not just as a clever mechanism to generate two-way ows. Second, over and above the predominance of ows between large industrialized countries, which can be explained using models of horizontal FDI, Markusen wants to understand why countries with very low wages do not attract investment even for fragments of the manufacturing process that seem to require only unskilled labour. He explains this using the unied knowledge capital model in Chapters 7 to 9, which allows for vertical fragmentation of production within a single rm but spread across two countries. By supposing that even when tasks using unskilled labour require some xed input of technical or managerial know-how from skilled labour, he is able to show that countries where skilled labour is extremely scarce will attract virtually no FDI of any kind. Third, although he declares the horizontal motive to appear more empirically relevant than vertical fragmentation of rm production across countries, he views the vertical aspect of the tradebased literatures knowledge capital model as important to generate the complementarities between trade and investment that have driven the enormous growth of intrarm trade. Finally, he considers the facts that multinationals are prevalent in industries intensive in R&D and skilled
r 2009 Blackwell Publishing Ltd
113
labour to be of central importance in his models. He eschews the idea of FDI ows being connected to endowments of physical capital as being unsupported in the data, focusing instead on skilled and unskilled labour. By assuming that foreign rms are better at screening potential takeover targets, the Razin and Sadka models described above replicate the oft-cited stylized fact that rms taken over by foreign investors are more productive than their domestic counterparts. In later chapters, the authors capitalize on their expertise in the public nance aspects of FDI to revisit the role of taxation in multinational rm behaviour. They offer a new stylized fact: source country corporate tax rates impact the discrete decision of whether to establish a foreign afliate, but not the magnitude of foreign investment ows, while host country corporate tax rates affect both this discrete decision and the magnitude of ows. They provide a straightforward theoretical intuition rationalizing this result. If the source countrys tax rate is sufciently high compared with a potential host country, then a rm will decide to establish a foreign afliate. The amount of production activity transferred (or attributed) to the afliate then clearly depends on how high the host countrys effective tax rate is, as the host countrys stance on deductions for depreciation, etc., affects the amount of capital invested by the parent in its afliate. The magnitude of these effects and the potential gain from their proposal of policy coordination is not clear, but in light of Markusens assertion that taxes appear to be of secondary importance (p. 6) among the determinants of FDI, perhaps could be protably discussed. The texts offer substantive but distinct roadmaps to help the reader explore any of these issues on her own through simulation and econometric analysis. Markusens text presents strategies to address non-linear programming (discrete choice) problems using GAMS computer software. Razin and Sadka rigorously examine the implications of a number of econometric specications when taking their models to data on aggregate FDI ows. In particular, they illustrate the advantages of using a Heckman selection model to avoid problems generated by country-pair observations with zero FDI ows. The Heckman model predicts the expected magnitude of FDI ows between two countries given that there is any ow at all. It avoids the bias that arises if one simply treats all of the country pairs with no ows as uninformative, missing observations. Silva and Tenreyro (2006) discuss why the Poisson quasi-maximum-likelihood (PQML) specication might be superior in these situations,10 but recent Monte Carlo experiments by Martin
10
The Poisson specication is useful when the data exhibit a large mean but a high proportion of zeros. The Poisson quasi-maximum likelihood approach can achieve consistent estimators even if the data are not Poisson distributed (see Cameron and Trivedi 1998, p. 668).
r 2009 Blackwell Publishing Ltd
114
Katheryn N. Russ
and Pham (2008) suggest that the Heckman selection method may perform much better when there are many zero observations, as is generally the case with large panel data sets. Because the issue appears to remain somewhat arguable, Razin and Sadkas Chapter 8, comparing different methodologies (though not the PQML), is particularly relevant at this point in time.
115
shocks are more prevalent attracts very few of the illiquidity-prone investors who prefer portfolio investment. Thus, Razin and Sadka offer a rigorous theoretical justication of why FDI is large as a proportion of total capital inows in countries where there is less transparency about rm productivity and more macroeconomic volatility. While attacking a different question, Razin and Sadkas depiction of the impact of asymmetric information on the choice of FDI versus portfolio investment is striking in its similarity to Markusens exposition of the impact of asymmetric information on the choice of FDI versus licensing. There is an incompleteness in the FDI/portfolio investment model: why cannot an investor about to purchase a controlling interest in a rm buy a put option to sell at a price above the markets sceptical lemon-loss price before the purchase takes place? There appears to be an unexploited arbitrage opportunity. Markusens problem is a bit different in form, but with a similar incompleteness. A rm considers setting up an ofce to sell its product abroad, but is uncertain about market conditions in the host country. It does not invest abroad if the size of the market will make the sunk cost involved in establishing its own afliate for marketing and distribution worthwhile. A local agent could be contracted to market and distribute the product, but would extract some of the producers prots in return, as there is an incentive to minimize its marketing efforts for a given payment. Because it cannot credibly stipulate the local agents effort level within the contract, the rm must pay the agent an additional information rent to induce him to exploit the markets full sales potential, which the agent knows but the rm can only guess. The agent acts as an arbitrageur, in a sense, insuring the rm (for an extra fee) against the possibility that the foreign market is too small to make investment in an afliate protable. An arbitrageur in the Razin and Sadka model could serve a similar role, even without superior information, insuring the potential direct investor against the possibility of a lemon-loss in the event of a liquidity shock, but receiving an extra rent because it knows the direct investors outside option involves the lemon-loss. If one imagines a dynamic extension, liquidity shocks are not altogether different from demand shocks that could act in a manner very similar to Markusens demand uncertainty. The case also illustrates an unexplored bridge between the micro-level and macro-nance approaches: where does the role of the investor stop and the rm begin? In the Markusen models, the rm makes investment decisions. In Razin and Sadkas world, the investor runs the rm, at times through a manager.11
11
As a further illustration, in the literature of open-economy macroeconomics, the consumer is the investor and the rms manager acts in the interest of the consumer (since the manager generally is given the consumers discount factor when prot maximizing, be it constant or stochastic).
r 2009 Blackwell Publishing Ltd
116
Katheryn N. Russ
This begs the question, should the rm be insuring itself by licensing an overseas sales agent, or should the direct investor be insuring herself by purchasing a put option? Either approach should result in the same behaviour on the part of the afliate, but they currently are split into two separate literatures. Further, in either model, a true arbitrageur in the form of a broker of a complete set of state-contingent bonds that allow investors to insure fully against uncertainty in market conditions (output sales or liquidity shocks) without paying these extra rents, could be welfare improving. Empirically, it is not clear to what extent rms and investors actually employ these three types of hedging techniques when engaging in FDI.
See also Russ (2007) for a discussion of risk sharing and the asymmetric volatility of shocks.
117
work of Antras et al. (2007) suggests that it is. Is rm behaviour associated with vertical fragmentation (inside or outside the rm) more effective than horizontal FDI as a hedge against country-specic shocks, or does it exacerbate their effect? A few papers, notably Bergin et al. (2008) and a series of papers by McLaren, are beginning to construct frameworks to answer these questions looking at vertical fragmentation outside the rm, but the specic characteristics of multinationals proposed in the texts discussed above knowledge capital and specialized screening, asymmetric information and internalization or controlling ownership, industrial structure, intra-industry versus inter-industry ows of FDI have not yet entered the discussion. Similarly, to the degree that there exists a relationship between the return on net foreign assets and exchange rate behaviour as suggested by the recent valuation effect literature originated by Gourinchas and Rey (2007), Razin and Sadkas liquidity model and Markusens model of endogenous markups are likely to have implications for the behaviour of exchange rates in a dynamic model. Lilia Cavallari (2007), for example, links multinational activity with deviations from purchasing power parity. Smith and Valderrama (2009) have begun to uncover the business cycle properties of FDI versus portfolio or equity investment in a small open economy in a model with perfect competition and a representative rm, but the business cycle implications of choices made by rms that behave as modelled by Markusen and direct investors challenged in the ways described by Razin and Sadka have yet to be explored. Antras and Helpman (2006) have pushed forward the micro-level theory of internalization is there a ip side relating to the investors choice of FDI versus portfolio investment? That is, are there benecial external economies for the agent investing in an MNE instead of purchasing securities from separate rms listed in separate markets, over and above any boost in protability arising from internalization? We know that MNEs act as nancial intermediaries in a direct way by channelling investment funds between branches in different countries. Does the internalization of diverse production and sales activities by the multinational rm also act as a form of nancial intermediation when it forces long-term cross-border investment due to Razin and Sadkas lemon-liquidation cost? There is also a question of whether FDI in different industries nontraded versus tradable, or nancial versus non-nancial, intermediate versus nal goods has different implications for a countrys business cycle. On the issue of nancial versus non-nancial industries, at least, Cetorelli and Goldberg (2008) and de Blas and Russ (2008) suggest this could well be true. If so, which market structures and which industries should be the focus of a unied theory of FDI in an open economy? Finally, if rms are earning
r 2009 Blackwell Publishing Ltd
118
Katheryn N. Russ
revenues in one currency and paying dividends in another, then Russ (2007) suggests that the source of nancing, omitted for the sake of clarity in past models, may motivate a complex relationship between FDI and exchange rate uctuations. Outside of a few papers like Froot and Stein (1991), the relevance of the source of nancing the country in which nancing originates is a topic with little empirical or theoretical guidance to date. In short, from fragmented literatures, a unied theory of FDI is slowly emerging. On the foundations of general equilibrium, sunk costs, asymmetric information and economies of scale, trade-based modelling engines are slowly merging with the analysis of capital ows to form a new theory of FDI. All of the questions mentioned above regarding risk sharing, business cycle transmission and behaviour, exchange rate behaviour, industryspecic considerations, and the structures and sources of nancing could have non-trivial implications for policy analysis as well as our basic understanding of the role of the multinational in the global economy. It is fortunate that the scholars who helped originate the literatures are providing the eld with manuals to guide the way forward. Despite profoundly different philosophical beginnings, they are leading us down converging paths. Katheryn N. Russ Department of Economics University of California, Davis One Shields Avenue Davis, CA 95616 USA knruss@ucdavis.edu
References
Aizenman, Joshua (1992), Exchange Rate Flexibility, Volatility, and Domestic and Foreign Direct Investment, IMF Staff Papers, 39(4), 890922. Antras, Pol, and Elhanan Helpman (2006), Contractual Frictions and Global Sourcing, NBER Working Paper No. 12747. Antras, Pol, Mihir A. Desai and C. Fritz Foley (2007), Multinational Firms, FDI Flows and Imperfect Capital Markets, NBER Working Paper No. 12855. Bergin, Paul R., Robert C. Feenstra and Gordon H. Hanson (2008), Outsourcing and Volatility, Manuscript, University of California, Davis. Cameron, A. Colin, and Pravin K. Trivedi (1998), Microeconometrics: Methods and Applications. New York: Cambridge University Press.
r 2009 Blackwell Publishing Ltd
119
Cavallari, Lilia (2007), A Macroeconomic Model of Entry with Exporters and Multinationals, Berkeley Electronic Journal of Macroeconomics, 7(1), 32. Cetorelli, Nicola, and Linda Goldberg (2008), Banking Globalization, Monetary Transmission, and the Lending Channel, Manuscript, Federal Reserve Bank of New York. Contessi, Silvio (2007), International Macroeconomic Dynamics, Endogenous Tradability, and Foreign Direct Investment with Heterogeneous Firms, Manuscript, Federal Reserve Bank of St. Louis. de Blas, Beatriz Perez and Katheryn N. Russ (2008), FDI in the Banking Sector: Why Lending Costs Fall when Spread Proxies Increase, Manuscript, University of California, Davis. Feenstra, Robert C. (ed.) (1983), Essays in International Economic Theory: Jagdish Bhagwati. Cambridge, MA: MIT Press. Feenstra, Robert C. (2004), Advanced International Trade: Theory and Evidence. Princeton, NJ: Princeton University Press. Froot, Kenneth A., and Jeremy C. Stein (1991), Exchange Rates and Foreign Direct Investment: An Imperfect Capital Markets Approach, Quarterly Journal of Economics, 106(4), 1191217. Gourinchas, Pierre-Olivier, and Helene Rey (2007), International Financial Adjustment, Journal of Political Economy, 115(4), 665703. Hymer, Stephen H. (1976), The International Operations of National Firms: A Study of Direct Foreign Investment. Cambridge, MA: MIT Press. Martin, William, and Cong S. Pham (2008), Estimating the Gravity Model when Zero Trade Flows Are Frequent, MPRA Paper No. 9453, University Library of Munich, Germany. Mundell, Robert A. (1957), International Trade and Factor Mobility, American Economic Review, 47(June), 32155. Ramondo, Natalia, and Veronica Rappaport (2008), The Role of Multinational Production in Cross-Country Risk Sharing, Manuscript, University of Texas at Austin. Russ, Katheryn N. (2007), The Endogeneity of the Exchange Rate as a Determinant of FDI, Journal of International Economics, 71(2), 34472. Silva, J. M. C. Santos, and Silvana Tenreyro (2006), The Log of Gravity, Review of Economics and Statistics, 88(4), 64158. Smith, Katherine A., and Diego Valderrama (2009), The Composition of Capital Inows when Emerging Economies Face Financing Constraints, Journal of Development Economics, in press.